What does it take to run a web app with 5K – 10K users?

Well, I think good indexing is key. How do you have your indexes. If they are aligned with what is frequently queried then you should be in good shape. A database of 5-10K users isn't difficult these days for any major DBMS. The key lies in the other things attached to the users and items you query. That can run into the millions. Even that is not much, but depending on how many times you query a million row table, it could put a pinch on your optimization.

Good database mechanics is key. That is the most important thing in my opinion. That is really the whole point in rails when you are deciding relationships. The abstraction in Rails when deciding what should be the best model structure is the same thing as deciding what should be the best and most efficient table structure in your database.

The rave about MongoDB is that it (maybe not quote me on this) "cures" the need for the desire multi-dimensional database. However, even with MontoDB's ability to expand due to it's not needing a pre-defined structure and the ability to expand out dimensionally to a certain extent, PostgreSQL (claims anyways) is still more efficient if you correctly index your tables (think about how you will be querying) and create the correct relationships. Build out models. Allow flexibility.

Also, don't forget caching. Redis and performing jobs is key in certain situations. However, don't get caught up in too much hype. Especially those coming from closed source technologies (Not just talking about caching technologies here but everything in general). They will sell and produce an atmosphere of necessity, but do some research first. Don't follow the heard. I am not going to call anyone out on this. Just do the research and think why is that necessary. I've mentioned Redis a few times and maybe that isn't even necessary either.

Most importantly, put your stuff out there. If it crashes, so what! At least you know you have something. And then you will have people who will give you advice in a coherent direction if necessary.

I salute you in your efforts. Now the most important part is put it out there and kick ass!

You may even want to look into Redis which is a cache system.

@DHH's startup school talk says it best.

"Scaling problems rule . . . "

http://www.youtube.com/watch?v=0CDXJ6bMkMY&t=23m34s

Sounds like you have the skills to grow this up to 500 to 1k paying customers on your own. Early on you can increase ram/cpu to handle any initial scaling issues.

Once you even have 1k customers you'll have revenue to hire experts to help you with scaling and security.

Good luck in 2015.

1. First off, don't think too much about what programming language/framework to use. For such a user base you can use mostly anything and it'll run smooth on a low spec server. Using Ruby and Rails for a website with 5-10k users (assuming they're active users) will definitely not cause problems or empty your CC.

2. Do not invest time and/or money in learning another programming language or framework until you are sure that for a specific component of your product, programming language X will perform at least 2 times better with 2 times less HW resources.

3. Stressing again on the app stack (I saw some really pushy comments on changing the programming language), it is rarely the bottleneck of a web app. You'll scale your storage stack way earlier and more often than the app stack.

4. Know your data. That's how you decide if it's better to use a RDBMS, document store, k/v store, graph database etc. Like I said before, you're going to scale your data before any other layer becomes a problem so choosing the right data storage solution is crucial. Don't be afraid to test various storage solutions. They usually have good -> great documentation and ruby tends to be a good friend to every technology. There's a gem for everything. :)

5. Scale proportionate to your business/product growth. You will have to scale at some point. But be careful to scale proportionate to your growth. For example, if the number of users will double, get the hardware that suffices that growth. Less HW resources will lead to a slower user experience thus user dissatisfaction. More HW resources than needed will increase your costs and the resources that are not needed will stay unused. Why waste money?!

These are my 2c. As your business gets bigger - I hope it does - other problems will occur. But usually these things will last up to 100k users.

Disclaimer: this is for a generic web app as you didn't give us any details. Depending on the app, some of my points might be inaccurate or invalid.

This is something I'm going to have to research. I haven't dealt with logging yet, besides reading development logs and Heroku logs.

Just a quick note: The links to foods on your front page point to the wrong URLs :)